Re: [INTERFACES] fe_setauthsvc: invalid name. Ignoring... ERROR - Mailing list pgsql-interfaces

From Tom Lane
Subject Re: [INTERFACES] fe_setauthsvc: invalid name. Ignoring... ERROR
Date
Msg-id 706.934813456@sss.pgh.pa.us
Whole thread Raw
In response to Re: [INTERFACES] fe_setauthsvc: invalid name. Ignoring... ERROR  ("Dutt H Kalapatapu" <dutt@eudoramail.com>)
List pgsql-interfaces
"Dutt H Kalapatapu" <dutt@eudoramail.com> writes:
> I finally figured out why I am gettting an authentication failure. The
> problem was my previous code has the declaration for PgDatabase like
> this.

> PgDatabase * myDatabase = new PgDatabase("dbname=template1");

> I was running this code on solaris and everything was working fine.  I
> looked at the PgDatabase class and figured out that we no longer need
> that 'dbname='.  Now my code looks like this:

> PgDatabase * myDatabase = new PgDatabase("template1");

>  and I could now run this from the command line.

Hmm, that explains a couple of things.  The first example (with dbname=)
is correct for Postgres 6.5 and later, because PgDatabase's constructor
now takes a conninfo string a la PQconnectdb.  Older releases used to
just take a database name, IIRC.  So, you are apparently using a
back-rev Postgres on your "new" setup.

I had looked at the code in fe_setauthsvc() and was unable to figure out
a reason for it to fail (it's almost dead code, anyway) --- but I was
looking at the current code.  I do seem to recall that pre-6.5 libpq++
would deliver this misleading error message on pretty much any failure
to connect, including specifying a wrong database name.

I'd recommend upgrading to 6.5.1 ASAP.
        regards, tom lane


pgsql-interfaces by date:

Previous
From: Peter Mount
Date:
Subject: Re: [INTERFACES] JDBC query...
Next
From: "Ken J. Wright"
Date:
Subject: Re: catching posting exceptions